If you have ever tried to book a hotel online and found yourself unsettled by an AI chatbot that seems to know just a little too much or not nearly enough, you are in good company. A new study from Texas A&M College of Agriculture and Life Sciences has confirmed what many travelers have long suspected: hotel booking chatbots are genuinely creeping people out, and that discomfort is directly hurting bookings. The research, which surveyed 340 adults in the United Kingdom who had used chatbots to secure hotel reservations, identified three primary sources of what they call the 'ick factor': inaccuracy, deceptive behavior, and intrusiveness.
The Root Causes of Chatbot Creepiness
According to the study, inaccuracy is by far the biggest offender. When a chatbot provides incorrect room rates, fumbles cancellation policies, or dodges straightforward questions, users react with far more than simple frustration. The negative emotional response triggered by inaccuracy was more than four times stronger than reactions to deceptive behavior or intrusiveness. This suggests that reliability is the bedrock of user trust in automated systems. A single error can feel like a betrayal, especially when the chatbot has been designed to sound friendly and capable.
Deceptive behavior, the second culprit, includes instances where the chatbot pretends to be human without disclosing its artificial nature. Many hotel booking platforms still deploy bots that use human names or avatars, and users who later realize they have been interacting with an AI feel manipulated. This deception amplifies the sense of unease, as it undermines the user's sense of control and transparency. Intrusiveness, the third factor, refers to chatbots that ask for excessive personal information early in the conversation, such as full credit card details or passport numbers, without explaining why it is needed. Users feel their privacy is being invaded, especially when the bot insists on continuing the interaction even after they have indicated a desire to stop.
The Uncanny Valley of Chatbots
The researchers also highlighted an effect known as the 'uncanny valley,' a concept originally used to describe humanoid robots that look almost but not quite human, causing a sense of revulsion. In the context of chatbots, the uncanny valley emerges when an AI tries too hard to mimic human conversation but fails at basic tasks. Lead researcher Babak Taheri explained that when a human-like system fails to actually behave like one, it triggers something deeper than disappointment in users. It can evoke feelings of unease, suspicion, and even distrust that go beyond rational evaluation of the bot's performance. For example, a chatbot that uses casual language and emojis but then gives a completely wrong check-in time is more unsettling than a simple, purely functional bot that might make the same mistake. The failure feels more jarring because the system promised a human connection but delivered a mechanical error.
Quantifying the Damage
The study provided concrete numbers to illustrate the business impact. The discomfort caused by these chatbot flaws cut users' willingness to continue chatting with the bot by nearly 38 percent. More critically, it nearly doubled the likelihood that users would delay or abandon their booking altogether. For the hotel industry, which relies heavily on online conversion rates, this is a significant loss. The researchers estimated that improving chatbot accuracy alone could recover a substantial portion of lost bookings, potentially worth millions of dollars globally.
A Simple Fix That Most Hotels Ignore
There is good news: the study also found a remarkably simple solution that most hotels are not currently using. When a chatbot clearly declares that it is an AI at the outset, users become far more forgiving of its mistakes. A simple opener such as 'Hi, I am your AI assistant' goes a long way toward managing expectations and reducing discomfort. The transparency seems to lower the threshold for tolerance; users understand that the system may not be perfect and adjust their trust accordingly. Yet many hotel booking platforms still avoid such disclosure, perhaps fearing that it will reduce the perceived quality of service. The research suggests the opposite is true: hiding the AI nature actually makes users more likely to feel deceived and abandon the interaction.
Along with transparency, the researchers recommend making it easier for users to reach a real human when their query becomes complex. A seamless handoff from bot to human agent can salvage a booking that might otherwise be lost to frustration. They also urge hotels to invest in upgrading the AI itself so that it can handle the basics without fumbling. The combination of clear AI disclosure, easy escalation paths, and improved accuracy can significantly boost user satisfaction and conversion rates.
Industry Context: The AI Travel Boom
This research lands at a fascinating moment for the travel industry. AI-powered travel booking is one of the hottest areas in tech right now. Google recently added AI trip planning capabilities to its search engine, allowing users to generate personalized itineraries through natural language queries. Uber also launched hotel booking capabilities within its app, powered by Expedia's inventory and AI-driven recommendations. These moves signal a broader shift toward integrating artificial intelligence into every step of the travel planning process, from flight searches to hotel bookings to itinerary management. But the Texas A&M study serves as a cautionary tale: implementing AI without careful attention to user psychology can backfire. The creepiness factor is not just a minor annoyance, it can actively repel customers.
The challenge for the hospitality industry is to design chatbots that are both helpful and trustworthy. This means not only improving the underlying natural language processing and data accuracy but also considering the human factors that shape user experience. Anthropomorphism, or giving non-human entities human-like qualities, can be a double-edged sword. While a friendly chatbot might feel more approachable, it also raises expectations that, when not met, lead to a deeper sense of betrayal. The uncanny valley effect is particularly pronounced in service contexts where errors have real financial and emotional consequences, such as booking a dream vacation or securing a hotel for a business trip.
Some hotel chains have already begun to experiment with AI disclosure and tone adjustments. Marriott, for example, uses a chatbot named 'ChatBotlr' that clearly identifies itself as a digital assistant. Hilton's 'Connie' robot concierge also introduces itself as a robot. However, many smaller or independent hotels still rely on generic chatbots from third-party platforms that do not prioritize transparency. The study suggests that a simple policy change—adding a line like 'I am an AI assistant' at the start of every chat—could yield significant improvements in customer satisfaction and booking rates.
Beyond the hotel industry, the implications of this research extend to any business using conversational AI for customer service. E-commerce sites, airlines, and even healthcare providers can learn from the findings. The key takeaway is that users are not inherently averse to interacting with AI; they are averse to feeling deceived, confused, or frustrated. Transparency, accuracy, and respect for user privacy are the foundations of a successful chatbot implementation. As AI becomes an ever more ubiquitous part of daily life, understanding the emotional and psychological impact of these interactions will be critical for maintaining consumer trust.
The Texas A&M study adds empirical weight to what many travelers have felt intuitively: that there is something genuinely unsettling about a chatbot that tries to act human but fails. By identifying the specific triggers of that unease and offering a straightforward solution, the research provides a roadmap for the industry to follow. The next time you book a hotel and a chatbot pops up, pay attention to how it introduces itself. If it says it is an AI, you might be more patient with its mistakes. If it tries to pretend it is human, you might want to brace for a creepy experience—and maybe call the front desk instead.
Source: Digital Trends News